Separation & Purification: Endotoxin Reduction Using Disposable Membrane Adsorption Technology in cGMP Manufacturing

Published on: 
BioPharm International, BioPharm International-05-01-2007, Volume 20, Issue 5
Pages: 40–45

Membrane-based chromatography technologies sometimes offer advantages over resin-based technologies.


The bioprocessing industry is increasing its use of membrane chromatography technology, especially for contaminant removal. The high flow rates, ease of use, disposability, and relatively low cost make it ideal for polishing applications. In a case study from Avecia Biotechnology, UK, the Sartobind Q (Sartorius) membrane chromatography product was used to remove endotoxin from a process stream. The result demonstrated advantages in time, product recovery, and cost over a traditional column step, leading to its use in a cGMP manufacturing process.

In biopharmaceutical manufacturing, it is essential that any products intended for injection into human subjects are produced virtually free of potentially harmful contaminants such as endotoxins and viruses. To ensure this, manufacturers incorporate certain contaminant removal steps into their processing procedures.

Endotoxins are typically associated with Gram negative bacteria, e.g., E. Coli. As the prefix "endo" suggests, this type of contaminant is not secreted by bacteria (as are exotoxins), but derives from the structure of the bacteria, specifically, from part of the outer monolayer of the outer membrane.1 When bacteria are lysed, whether intentionally or as a side effect of the process, they release these structural components.

Introducing even small amounts of certain endotoxins (e.g., lipopolysaccharides or LPS) into the bloodstream can cause a severe immune reaction. The presence of endotoxins triggers the immune system's signaling cascade, leading to the secretion of cytokines. The subject develops an abnormally high body temperature and respiration rate and low blood pressure. This can lead to endotoxic shock, which may be fatal. Therefore, it is essential to develop a robust process step to remove endotoxins from the process stream.

One mode of separating the target substance from endotoxin molecules is by size exclusion via ultrafiltration. But this requires a significant difference in size between the target molecule and the endotoxin, as when the target substance is a nucleotide. In this case, the smaller target molecule passes through the membrane while the endotoxin is retained. Often, however, the target molecule and the endotoxin molecule fall into a similar size range, so separation based on size exclusion is not applicable. The size of endotoxin molecules can range from <10 kDa in a monomeric form to >10,000 kDa in an aggregated form.2

An alternative mode of endotoxin removal is separation based on charge. This is possible because of the strong negative charge associated with endotoxin molecules under conditions around pH 8.0. This charged state is a result of the ethanolamine part of the endotoxin molecule having no charge at this pH, whereas the lipid-A and core polysaccharide regions are negatively charged at pH 8.0.

Column chromatography is commonly used to take advantage of charge as a tool for separating proteins. In this process, a micro-porous stationary phase, conventionally a gel or resin consisting of agarose or cellulose beads, is packed into a tube or column of glass or stainless steel. By coating the stationary phase with anionic groups (ion-exchange chromatography), it is possible to selectively adsorb certain charged components of a process stream, depending on the conditions of the mobile phase. For example, under the conditions of low conductivity (<20 mS/cm-1 ) and pH around 8.0, certain molecules (including endotoxin) have a net negative charge, and therefore, will bind to a positively charged stationary phase. Thus, they are retained in the column (removing them from the process stream), while the non-binding (target substance) molecules pass through the column. The outcome is a significant reduction of endotoxin levels in the process stream.

Such resin and column-based chromatography systems, however, suffer from drawbacks such as low flow rates, the need to pack the column with stationary phase, long regeneration times, channeling, and susceptibility to fouling.3 In the context of contaminant removal (i.e., binding impurities rather than binding target proteins), these drawbacks become more of an issue. Because contaminant removal is usually a "polishing" step, flow rates and convenience are priorities, and these are not strengths for resin or column chromatography. Even with relatively low levels of contaminant in the process stream, low flow rates mean that conventional resin-based columns might need to be quite large, thus further raising costs.4


To overcome these disadvantages, process specialists have been focusing increasing attention on the developing area of membrane-based chromatography. Membrane adsorbers, as they are often known, work on exactly the same principles as resin-based chromatography systems, with the ligand (e.g., quaternary ammonium) being bound to a support medium. In principle, any ligand that can be bound to a chromatography resin can be bound to a support medium, in this case a membrane.

One major difference between a membrane support and a resin support is porosity. Membrane adsorbers in ion-exchange membranes typically have a pore size of >3 μm. This nearly eliminates the diffusion limitation, improving flow rates and chromatographic kinetics. The near absence of pore diffusion means that with membrane-based ion-exchange modules, there is a significant level of flexibility in flow rate selection.

The open structure of the membrane is obvious when visually compared with a conventional bead of approximately 90 μm in size (Figure 1).

Figure 1

Because membrane chromatography capsules are single-use, they are simple to use. Once the module is installed and flushed with buffer, processing can begin. Once processing is completed, the module is removed and disposed of, without the need for washing, eluting, cleaning, or regenerating. This yields a significant reduction in processing time and buffer volumes.

With the increasing emphasis on reducing development times and costs, membrane chromatography products are being used in more and more cGMP processes.


The UK contract manufacturer Avecia Biotechnology successfully applied membrane adsorption in a cGMP process involving the manufacture of B2365, a recombinant protein produced in E. coli. The B2365 protein is initially expressed as a fusion protein and purified from the bacterial lysate by affinity chromatography. The fusion protein is then immobilized onto glutathione sepharose affinity chromatography media. The column is subsequently washed to remove the bulk of the impurities. Once washed, the immobilized glutathione-S-transferase (GST)-B2365 protein is cleaved from the column using a GST-3C protease, which also binds to the column, allowing the cleaved B2365 molecule to be washed from the column, leaving both the GST tag and GST-3C protease immobilized on the column. Although B2365 is of high purity after this single step, it is then further purified in a polishing chromatography step to remove free GST that may have leached from the column.


Previously, Avecia had purchased the GST-3C enzyme from an outside vendor, but to reduce overhead costs and to be able to certify the origins of the protease as non-animal-derived and free of transmissible spongiform encephalopathies (TSE), an internal project arose with the remit of producing 400 g of GST-3C protease for subsequent B2365 manufacture. GST-3C protease is a genetically engineered fusion protein expressed in E. coli, consisting of Mr 20,000 human rhinovirus 3C protease coupled with a Mr 26,000 GST tag. This protease was specifically designed to facilitate removal of the protease by allowing simultaneous protease immobilization and cleavage of GST fusion proteins on glutathione sepharose chromatography media. GST-3C protease specifically cleaves between the Gln and Gly residues of the recognition sequence of LeuGluValLeuPheGln/GlyPro.

Exploiting the GST tag for purification using glutathione sepharose affinity chromatography resulted in rapid purification and high purity levels after a single purification step. This constituted a quick and easy purification strategy, but analysis of the column eluate indicated, not unexpectedly, that the endotoxin level was extremely high.


A number of options were explored to reduce the endotoxin level, including anion exchange column chromatography and membrane chromatography. After small-scale laboratory work it was quickly determined that membrane chromatography technology would offer significant advantage with regards to product recovery, time saving, and equipment requirements during manufacture. A small-scale trial was performed using a Sartobind MA75 (2.1 mL bed volume) membrane adsorber (Sartorius). Under these conditions the membrane adsorber provided good performance in terms of product recovery and product flow rates. From the data generated the membrane was scaled-up for use within a 100 L scale process to a membrane bed volume of 70 mL, or a 5-in. capsule.

Before GMP manufacture, a 100 L process demonstration batch was performed to assess the feasibility of the process. First, the fermentation broth was high-speed centrifuged to separate the cells from the basal media. The cell paste was then resuspended in a controlled buffered solution before high-pressure homogenization. The bacterial lysate was then 0.22-μm filtered before application to the chromatography column.

A sample of the column eluate was taken for endotoxin analysis, giving a value of 42,900 EU/mL. The bulk-process solution was then passed through the membrane adsorption capsule with further samples being taken at 5-L intervals for endotoxin analysis to model the capsule's clearing capacity. Once passed through the capsule, the GST-3C protease was then concentrated to 5 g/L and diafiltered into a final formulation buffer. The final product was then filtered prior to long- term storage at -70 °C. The endotoxin removal profile is shown in Figure 2.

Figure 2

Using the data obtained from the process demonstration batch, the membrane adsorber capsule was resized prior to the GMP manufacture to a 20-in. (360-mL bed volume) capsule. During November 2005, two 100-L cGMP batches were performed under the same processing conditions as the process demonstration batch, except that the column eluate was then passed through a 20-in. Sartobind SingleSep (single use) capsule. The results from the endotoxin removal step are shown in Table 1.

Table 1. Data from the Sartobind Membrane chromatography process step from two separate batches


The use of disposable-membrane adsorption technology instead of traditional column chromatography greatly reduced plant preparation time, because the requirement for column packing, assessment, and sanitization, was removed, along with the associated documentation, equipment, and QC testing required to perform column chromatography. From the two batches of the cGMP manufacturing project, a total of 415 g GST-3C protease was produced, with significant cost savings. Final product analysis can be seen in Table 2.

Table 2. Batch data from the final product, from two separate batches

Andrew Clutterbuck is a senior purification scientist at Eden Biodesign; James Kenworthy is a purification specialist at Sartorius Biotechnology, Ltd., +44. (0).1372.737.159, and John Liddell is head of process science at Avecia.


1. Alberts B et al. Molecular biology of the cell. Garland Publishing, Inc., New York: 1983.

2. Belanich M et al. Reduction of endotoxin in a protein mixture using strong anion-exchange membrane absorption. Pharm Technol. 1996; March;142–5.

3. The Sartorius group. Endotoxin removal, ion exchange chromatography with Sartobind membrane adsorbers. Application note. Gottingen, Germany: Sartorius.

4. Gottschalk U, Fischer-Früehholz S, Reif, O. Membrane adsorbers: a cutting edge process technology at the threshold. Bioprocess Int. 2004; May;2(5):56–65.

5. Fischer-Früehholz, S. Membrane adsorbers part 1: new dimensions in chromatographic purification. GIT Lab J. 2005;2:53 –5.